def _handle_target_table_token(self, sub_token: TokenList) -> None:
if isinstance(sub_token, Function):
# insert into tab (col1, col2) values (val1, val2); Here tab (col1, col2) will be parsed as Function
# referring https://github.com/andialbrecht/sqlparse/issues/483 for further information
if not isinstance(sub_token.token_first(skip_cm=True), Identifier):
raise SQLLineageException(
"An Identifier is expected, got %s[value: %s] instead" %
(type(sub_token).__name__, sub_token))
self._lineage_result.write.add(
Table.create(sub_token.token_first(skip_cm=True)))
elif isinstance(sub_token, Comparison):
# create table tab1 like tab2, tab1 like tab2 will be parsed as Comparison
# referring https://github.com/andialbrecht/sqlparse/issues/543 for further information
if not (isinstance(sub_token.left, Identifier)
and isinstance(sub_token.right, Identifier)):
raise SQLLineageException(
"An Identifier is expected, got %s[value: %s] instead" %
(type(sub_token).__name__, sub_token))
self._lineage_result.write.add(Table.create(sub_token.left))
self._lineage_result.read.add(Table.create(sub_token.right))
else:
if not isinstance(sub_token, Identifier):
raise SQLLineageException(
"An Identifier is expected, got %s[value: %s] instead" %
(type(sub_token).__name__, sub_token))
self._lineage_result.write.add(Table.create(sub_token))
def _handle_source_table_token(self, sub_token: TokenList) -> None:
if isinstance(sub_token, Identifier):
if isinstance(sub_token.token_first(skip_cm=True), Parenthesis):
# SELECT col1 FROM (SELECT col2 FROM tab1) dt, the subquery will be parsed as Identifier
# and this Identifier's get_real_name method would return alias name dt
# referring https://github.com/andialbrecht/sqlparse/issues/218 for further information
pass
else:
self._lineage_result.read.add(Table.create(sub_token))
elif isinstance(sub_token, IdentifierList):
# This is to support join in ANSI-89 syntax
for token in sub_token.tokens:
if isinstance(token, Identifier):
self._lineage_result.read.add(Table.create(token))
elif isinstance(sub_token, Parenthesis):
# SELECT col1 FROM (SELECT col2 FROM tab1), the subquery will be parsed as Parenthesis
# This syntax without alias for subquery is invalid in MySQL, while valid for SparkSQL
pass
else:
raise SQLLineageException(
"An Identifier is expected, got %s[value: %s] instead" %
(type(sub_token).__name__, sub_token))
